After a disaster, such as an earthquake or hurricane, victims are often trapped in voids that are survivable but too difficult or dangerous for rescuers to reach. Small robots that can crawl through narrow crevices could provide rescuers with valuable information about where to focus their efforts. Although individual small-scale robots have previously lacked the intelligence and mobility of larger robots, recent breakthroughs in robot locomotion, and compact computers (from cell-phone technology) will enable the development of distributed robotic teams of dozens of very low-cost, small-scale mobile robots coordinating with each other to overcome obstacles while rapidly making a map of an irregular environment. This National Robotics Initiative (NRI) award supports fundamental research in understanding how to design systems of multiple robots which can cooperate with each other to identify and then move over obstacles, while providing information to, and receiving overall guidance from, human operators. Success in this research project will bring society closer to having teams of mobile, disposable, search and rescue robots which can robustly locomote through uncertain environments and find survivors in disaster situations while removing risks posed to human and animal rescuers. In addition to graduate students, undergraduate students will gain firsthand experience in the design and control of millirobotic platforms, and robot designs will be made publically available for construction and simulation. To increase youth interest in science, technology, engineering and math, middle school students will be engaged in robot design and build activities.
This research project aims to understand how to combine enhanced mobility through dynamic robot cooperation with distributed low-cost mapping of irregular environments. To reduce cost and complexity, millirobots are typically minimally actuated and have limited mobility in complex terrain. Through dynamic robot cooperation, multiple robots can jointly traverse terrain features which can block single robots. Algorithms will be developed to flexibly surmount obstacles by combining forces from multiple robots and using robot body surfaces as temporary bridging or stepping elements, which should enable teams of millirobots to cooperatively locomote and explore areas efficiently, for example by finding, or if necessary, creating low-energy ingress paths. The research team will develop new techniques for joint 3D mapping using multiple small and medium robots size equipped with lightweight sensors. A research challenge is using compact (sub-5 gram) steerable lasers and image sensors to identify reference points, and then using low-bandwidth communication of the most salient features to higher level computation nodes to create a 3D global map. Exterior surface maps will be fused with interior maps to guide search planning towards the regions with highest probability of unexplored volumes. Human searchers can carry lightweight mapping gear (on the order of 100 grams) while working on exterior surfaces, and provide information regarding the search regions. The robot ensemble will be tested at a local urban search and rescue training facility.
